Customizing mass production
Database marketing @ Organon
Alexander Ivo Thiadens 27 December ’02
Customizing mass production
An online market research of a psychiatrists' virtual communityAuthor Alexander Ivo Thiadens
0978590
Mentor at RuG Drs. R.A.J.M. Valens Dr. W. Jager
“The old paradigm, a system of mass production, mass media, and mass marketing
is being replaced by a totally new paradigm, a one-to-one economic system. The 1:1 future will be characterized by customized production, individually addressable media and 1:1 marketing totally changing the rules of business competition and growth.”
Preface
From the young years of my life I am fascinated by the possibilities of the fast development, that computer technology brings into the rules of business. Especially during my study Industrial Engineering and Management Science at the University of Groningen, I have discovered the impact of ICT in the traditional business. I have developed a strong interest in the combination of modern ICT and the Marketing at the World Wide Web. With the Internet, a full worthy channel for communication between customer and producer did arise. It is my opinion that the Internet has even broader possibilities for convincing a consumer than the traditional channels. New database technology opens the possibilities of customer need mapping on the Internet and no technology and information push anymore. The fast development of the data interchange technologies will achieve a full interactive Internet, as it almost is these days. I want to underline once more that the Internet is not a replacement of the traditional channel, like the new economy disaster proved. However, it is a fulfillment, an extra channel in the multi channeling approach.
In this internship and the research I conducted at Organon, I want to learn and discover the e-marketing strategies of today. I want to explore the contradiction the ICT started…….customizing mass production.
I hope you will enjoy reading this thesis the same as I did enjoy my internship at Organon’s e-business department.
Alexander Ivo Thiadens
Contents
Chapter 1 - Customizing mass production @ Organon ... 6
1.1. Reason for the research... 7
1.2. Research Type ... 9
1.3. Problem definition... 9
1.3.1. The research goal ... 9
1.3.2. Problem question... 10
1.3.3. Research preconditions... 10
1.4. Specified research questions ... 10
1.5. Research Draft ... 12
1.6. Research approach ... 12
1.7. The organization of this thesis... 14
Chapter 2 – theoretical framework ... 15
2.1. Commercial virtual community ... 15
2.1.1. Stickiness... 16
2.2. Defining the customer... 17
2.3. What is marketing intelligence?... 20
Chapter 3 - The environment of PsychiatryMatters.MD... 24
3.1. Organon NV ... 24
3.1.1. E-business department ... 26
3.2.1. Strategic business objectives ... 27
3.2.2. Organon’s Internet marketing objectives ... 29
3.2.3. Objectives PsychiatryMatters.MD ... 29
3.3. Online customer group... 30
3.4. Summarizing the chapter ... 31
Chapter 4 - Design of the research ... 32
4.1. In coming data sources ... 32
4.1.1. Data mining ... 34
4.2. The user table as population framework ... 36
4.3. Origination of the user... 36
4.3.1.Method of sample taking... 38
4.4. Clustering the users... 38
4.4.1. Cluster algorithm... 40
4.5. Goals that will be measured ... 41
4.6. Summarizing the chapter ... 42
Chapter 5 - Results of research ... 44
5.1. The user table as population framework ... 44
5.2. Origination of the subscribers... 45
5.2.1. Search engines... 46
5.2.2. Search strings... 47
5.3. Segmenting the subscribers... 49
5.3.1. Naming segment 1... 51
5.3.2. Naming segment 2... 53
5.3.3. Naming segment 3... 54
5.3.4. Naming segment 4... 56
5.3.5. Naming segment 5... 57
5.4. Summarizing the chapter ... 58
Chapter 6 - Conclusions and recommendations ... 59
6.1. Origination of the subscribers... 59
6.1.1. Search engines... 59
6.1.2. Search strings... 60
6.2. Segmenting the subscribers... 60
6.2.1. User portfolio analysis... 62
6.3. Recommendation – What to do with this intelligence? ... 63
6.3.1. Checking the goals of PsychiatryMatters.MD ... 64
6.4. Summarizing the chapter ... 66
Chapter 1 - Customizing mass production @ Organon
The traditional mass industry did use product-based production lines and more important in this approach the customer sits third row or even worse. The modern information and communication technology did change the traditional industry. These days production is adapted to customer demand and with this marketing has changed – dramatically. Once focused on products and the need to identify and exploit product-specific ‘gaps’ and advantages, successful marketers now must focus on customers, positioning products and customizing services to the target markets and satisfy widely disparate market niches.
We continue to see more and more marketers strive to improve performance by using consumer behavior, demographics and lifestyle information to not only segment the marketplace, but also to orchestre highly targeted marketing communication strategies.
We continue to see more emphasis on customer service and customer loyalty programs as ways to strengthen and sustain customer relationships in today’s crowded and highly competitive marketplace. Also more and more companies demassify their marketing efforts in an attempt to more effectively and efficiently achieve their sales and profit goals.
Interactive marketing, relationship marketing, database marketing, new direct marketing…. All try to name this development in the marketing today. All are information driven marketing processes, made possible by database technology that enables marketers to develop, test, implement, measure and appropriately modify customized marketing programs and strategy.
Recently Organon is using the Internet channel for focus on customers and to position pharmaceutical products. Organon is well known in the anti-conceptive market. However, smaller is Organon’s fame and market share for its anti-depressants. The e-business department did start initiatives for creating customer loyalty through specialized online portals. With the portal, “PsychiatryMatters.MD” Organon is hoping to create brand awareness and loyalty among prescribers towards their antidepressants. The following sections clarify the reason and the design of this research.
1.1. Reason for the research
The web portal created by Organon for psychiatrists is named PsychiatryMatters.MD. The portal is officially online since March 2001. Usually Portal is used as a marketing term to describe a web site that is or is intended to be the first place people see when using the Web. Typically, a portal-site has a catalogue of web sites, a search engine, or both. A portal site may offer email and other services to entice people to use that site as their main "point of entry" to the Web. In this specific case, the portal must be the point of entry to the brand sites of Organon. Appendix A shows a sitemap of www.PsychiatryMatters.MD.
With the start of the portal PsychiatryMatters.MD, the marketing management of Organon wants to reach its online customers. The main objectives set for the launch of this portal:
• Create/Improve/Support image Organon in psychiatry continuously;
• Obtain information about behavior of the target group.
The target group for the portal is all medical professionals with interest in psychiatry. Section 4.4. will specify the objectives. The end goal is to accomplish conversion with the subscribers and to turn non-prescribers into prescribers of Organon’s Central
Nerve System (CNS) products. Conversion in this case means guiding the target group through the portal site to the brand sites of Organon’s products.
New projects of Organon like Remeron that have been launched in unfamiliar or familiar markets never start of with an optimal functioning or optimal positioning in this market. Setbacks will arise due to unknown factors or due to unexpected changes in the market. Only a continuous flow of accurate market information to the marketing management of Organon will help solve problems cross the projects’ objectives. Chapter 2 will explain more about marketing intelligence.
To let the portal succeed in its objectives a fluently subscription of new visitors is an important aim. Therefore, the subscriber ratio at PsychiatryMatters.MD is an example of a subject that gets extra attention in this research to check the inflow of new subscribers. Another problem is that within the group that subscribes a major part does not belong to the target group. Does the portal attract new prescribers or are these new users non-targets? For this reason the research determines the characteristic origination at the Internet for the target group. In addition, do these new users keep interest in PsychiatryMatters.MD? For this reason this research monitors the stickiness of the site regarding the target group. Another problem is the uncertainty if the target group is really the target group. Are people pretending being a Medical professional? For this reason the user profiles of the target group will be compared with an independent representative database. The management wants to know if this project is effective and thereby the budget spent well. According to the objectives, it is better to spent one dollar for attracting one medical professional than one dollar spent on two consumers. Thus, it is not only the quantity that is important but also especially the quality of the customer group. Section 2.3. defines the customer group.
In this research, the most important aim is to track the customer group of PsychiatryMatters.MD during a period to find differentiating characteristics among the target group. First the characteristic origination before they log on to PsychiatryMatters.MD. Second and more important this research tries to determine differentiating behavior at PsychiatryMatters.MD. The results of this online market analysis will answer the questions raised above. An extra feature in this research is to check if the behavior is compatible with the objectives set with the launch of the portal ‘PsychiatryMatters.MD’. The final chapter will recommend future strategies or adaptations to the marketing campaigns.
1.2. Research Type
For the benefit of the use of PsychiatryMatters.MD in the professional world of psychiatry, this research reveals the behavior of the users. The portal supports the corporate image of Organon among the medical professionals and thereby influences the sales of CNS products.
The problem owner is the managing department on behalf of Organon for PsychiatryMatters.MD. In this, case the Internet marketing management in the e-business department. Therefore the proper characterization of this analysis is both a customer related field research and a business-policy-supporting research. In addition, this analysis is a problem-solving research [3]. Because a complicated data mining process makes it possible to analyze, interpret and state conclusions about the population. Section 4.1.1. explains the process of data mining in this research.
1.3. Problem definition
For a clear understanding of this research the research problem is divided into three parts, namely [17]:
• The research goal;
• The problem question;
• The preconditions.
1.3.1. The research goal
To check if the portal ‘PsychiatryMatters.MD’ reaches the target group marked out by the objectives set for the positioning in the professional psychiatry and identify differentiating characteristics of the target group.
An online market research of the users of the PsychiatryMatters.MD portal will create the necessary market intelligence. Two stages cover the research questions raised with the research goal. The first stage determines the characteristic origination of the target group. The second stage of this research starts with a cluster analysis. Cluster analysis is based on characteristics of customers in a way that the response on
market instruments within the segments is homogenous and between the segments is heterogeneous [11]. During four months, a tracking process of the segments reveals the characteristic behavior of the clusters. In the end this research compares the results of the analyses with the management objectives of this portal. If the market movements and the objectives are divergent, then the site or campaigns can be adapted or specified. Chapter 4 explains the methods of analysis.
1.3.2. Problem question
What are the specific characteristics of the segments within the target group of ‘PsychiatryMatters.MD’ and are they compatible with the objectives set by the management?
1.3.3. Research preconditions
Two main conditions divide the preconditions for this research. First, the preconditions that are related to the research process are distinguished. The market of PsychiatryMatters.MD’s users marks out the research population. Moreover, the research must hold scientifically approved methods of analysis.
Second, the preconditions that are related to the outcome of this research are distinguished. The research has to be finished in the end of the internship. Of course, the outcome must support the e-marketing management of Organon in their decision process.
1.4. Specified research questions
To answer the main research question (1.3.2.) it is easier to separate it in questions that are more specific. The answers to the specific research questions base the final findings of this research and the answer to the main research question:
1. To understand the purpose of the portal we have to know what the objectives
What is the Organon e-business organization in the environment of PsychiatryMatters.MD?
What are the measurable business goals set by Organon e-business for the portal PsychiatryMatters.MD?
2. For knowing what kind of customers we are looking for at PsychiatryMatters.MD:
What is the target group within the objectives set for the web portal PsychiatryMatters.MD by the e-marketing management?
3. The population framework that is used for the research must reliable. So, to check the reliability of the user profiles that are created by the users of PsychiatryMatters.MD themselves:
Are all users that are psychiatrists according to their user profiles truly psychiatrists?
4. For identifying the characteristics of the target group among the whole customer
group at PsychiatryMatters.MD:
What is the characteristic origination of the target group before they log on to PsychiatryMatters.MD? Phase I in the research.
What are the specific characteristics of the target group of PsychiatryMatters.MD after they log on to PsychiatryMatters.MD? Phase II in the research.
5. Finally, to compare the research results with the management objectives:
Are the characteristics of the target group found at PsychiatryMatters.MD compatible with the business objectives?
1.5. Research Draft
Figure 1.2. indicates the research draft that represents schematically the relations of the research object with relevant elements in its environment. In this case, the research object is the portal PsychiatryMatters.MD i.e. the system that is managed (MS) [3]. The elements influencing the portal are the objectives, the customer group and the managing department i.e. the managing organ (MO). Chapter 3 explains more extensively this environment of PsychiatryMatters.MD.
1.6. Research approach
This section describes the approach of this research.
1. First stage is all about describing the environment of PsychiatryMatters.MD. The research draft in chapter 1.5. indicates three objects in the environment that: the organization of Organon e-business, its business objectives and the customer
MS
Subscribers
Target group Other customers
Objectives
eBusiness MO
group. Chapter 3 describes these three objects because they influence the portal directly.
This stage uses information acquired through internal information sources like the marketing plan and external information sources like VODW Marketing and Synergetics.
2. For this research to be representative it must use a reliable population framework [1]. In this research the user table of PsychiatryMatters.MD The user table includes the customer profiles of the users of PsychiatryMatters.MD. This research compares the customer profiles should be compared first with an independent database to check the profiles for falsehoods.
3. The data is acquired through analysis of the actual online behavior of the users at PsychiatryMatters.MD during four successive months. This analysis has two stages:
a) A profile analysis identifies the origination of the subscribers of PsychiatryMatters.MD through the search engines and the search strings that the subscribers use. The data is extracted from the log files and the user table.
b) A segmentation analysis tracks characteristic behavior of the subscribers. A cluster analysis identifies several homogeneous segments with distinctive characteristics.
Both the analyses use data extracted from the log files and the user table. Section 4.1. describes the data sources and the data mining process.
4. The clusters will be tracked for four successive months to find out if the portal PsychiatryMatters.MD has ‘stickiness’. Chapter 4 explains why this research uses stickiness to name user retention. A tracking process identifies distinctive characteristics in the behavior of the clusters.
5. The results of the analysis will be compared with the management objectives to check if PsychiatryMatters.MD has the right target.
1.7. The organization of this thesis
This thesis uses the same steps as indicated in section 1.6. In this way it is easy to set recommendations and conclusions in the final chapter.
The first chapter describes and motivates this research. This chapter explains the problem definition and the research approach to start this research in the right direction.
The second chapter explains the theoretical framework. The theoretical framework sets the foundation of this research. The chapter explains the commercial virtual community like PsychiatryMatters.MD actually is. In addition, this chapter explains the necessity of marketing information behind PsychiatryMatters.MD for Organon. Further the chapter explains all the definitions and models.
The third chapter describes the environment of the organization of Organon as indicated in the research draft. Therefore, the chapter defines the organization of Organon e-business, its business objectives and the chapter defines the customer group.
The fourth chapter explains the design of the research. Thus, the explanation of both the analysis methods for the two stages in this research is set. First the chapter describes the checking of the user profiles for the user table to be a reliable population framework. Second, the chapter describes the data sources and the data mining process. In the end, the chapter describes the analysis methods for determining the characteristic origination and customer segments.
The fifth chapter describes the results of both the analyses. This chapter uses the same two-stage division as chapter four. First, a section describes the results on the origination analysis. Second, a section describes the results on the cluster analysis.
The final chapter states the conclusions and recommendations for the Organon management.
Chapter 2 – theoretical framework
This chapter sets the theoretical framework for the validation of this research. This research considers the portal PsychiatryMatters.MD a Commercial Virtual Community (CVC).
The first section 2.1. explains the CVC and definitions regarding the CVC will be set. Virtual communities have potential value for Organon. A CVC can represent a medium for interaction to take place and a source of market intelligence. The successive sections will explain this.
First section 2.2. explains the target group of PsychiatryMatters.MD. and will describe the funnel method.
Second section 2.3. describes the marketing intelligence extracted from PsychiatryMatters.MD.
2.1. Commercial virtual community
This research considers the portal PsychiatryMatters.MD a Commercial Virtual Community (CVC). At Organon PsychiatryMatters.MD is named a portal, but it has the same objectives and features as a CVC. The CVC takes part in the marketing strategy of Organon. This section explains a Commercial Virtual Community (CVC) and its role in the marketing strategy of Organon.
Among all kind of businesses, virtual communities are becoming very popular. The main reason behind their popularity is that customers spend a lot of time in these virtual environments. The goal of Organon for creating a CVC is the design of a new kind of culture to stimulate community life among people who might not have this ability in the physical world. A CVC can become a pool consisting of customers, which can benefit Organon [6]. The definition for a CVC in this research is:
“Commercial Virtual Communities are affiliate groups of people emerging on the Internet having a distinctive focus with commercial orientation, which have a specific membership focus and provide content to its participants. “[18]
Very important regarding a CVC is the retention of users [18]. Internet marketing and this research use the term 'stickiness' for describing the retention of users. The next section explains and defines the concept of stickiness.
2.1.1. Stickiness
PsychiatryMatters.MD only succeeds when it has the ability to attract and retain visitors. This chapter explains stickiness at a CVC.
Within academic literature, several authors define the concept stickiness. When comparing these various definitions, one aspect appears to be the most important. Namely the frequency of a user's visits determines the stickiness of a site. In this research, a site has stickiness if the user visits the site more than once. This research uses Davenport’s [2] definition of stickiness:
“Stickiness is a term used to describe a site’s ability to attract and retain visitors”
Thus, stickiness is a feature of a web site that results in increased frequency of a user’s revisits. Consequently, the drivers of stickiness are the components of a site that result in increased frequency of user’s revisits. The research of Walczuch, Verkuijlen, Geus and Ronnen [18] reports extensively on the drivers of stickiness. Table 2.1. shows a summery of the main drivers of stickiness.
Sticky type of Content Brief explanation
E-mail Free e-mail services for users Information e.g. articles, news, etc. Storage of information Users can store personal info Entertainment e.g. games, polls, etc.
Community Platform where users chat and discuss, e.g. chat, guest books Commercial services Services with a commercial
character offered to community members, e.g. online shopping
Sticky dimensions of content
Breadth of content A wide range of content available Depth of content Detailed content
Frequent updates Content is updated frequently Table 2.1.
A CVC always is set up for a target market. In this research it is very important to have clear understanding of the target group of PsychiatryMatters.MD. The next section defines the target user for PsychiatryMatters.MD, the needs and the funnel method to catch this customer on the Internet.
2.2. Defining the customer
The target group of PsychiatryMatters.MD is all medical professionals with an interest in psychiatry. The target group is the group of users that are most wanted by Organon at PsychiatryMatters.MD. The target group is the most profitable group of users in a loyal relationship with Organon. The customer group of PsychiatryMatters.MD is the actual user group. This customer group can include users that not directly influence sales of Organon’ products.
Important is to pull customers to PsychiatryMatters.MD. To do so it is important that the services and content of PsychiatryMatters.MD are adapted to the needs of the target group. A CVC becomes sticky to a user when it addresses certain basic needs of individuals. A theory that explains why individuals choose to use media in general and certain media channels in particular is the Media System Dependency (MSD) theory [4]. Therefore the reasoning in this section is based on the MSD theory. MSD theory argues that dependency relationships exist between individuals and media channels. People have certain basic needs and are dependent on the media channels to address them. Dependency exists because media controls most of our sources of information [4]. A CVC can be considered a media channel. Similar to TV, Radio and the newspaper, it also provides its users with a broad range of information. So relating this to stickiness, if a CVC is able to address the basic needs of an individual, it becomes sticky to that particular person. DeFleur and Ball-Rokeach [4] present a categorization of the general needs individuals have. These needs are understanding, orientation and play. The needs both relate to the individual himself and to his environment. Table 2.2. on the next page shows a brief overview of this categorization. Intelligence about the profile of a customer or a customer segment is decisive for Organon’s success. This intelligence is used to adapt services at PsychiatryMatters.MD more efficiently to the most desirable users and to reengineer the unprofitable services.
Understanding Orientation Play
Self Self-understanding
e.g. learning about oneself and growing as a person.
Action orientation
e.g. deciding what to buy or how to dress.
Solitary play
e.g. relaxing when alone.
Social Social understanding
e.g. knowing about and interpreting the world of community..
Interaction orientation
e.g. getting hints on how to handle new or difficult situations Social play e.g. going to a movie or chatting on the Internet with colleagues. Table 2.2.
The services of PsychiatryMatters.MD are adapted to customer profiles to trigger the revisits of the profitable users. In this case of PsychiatryMatters.MD the sticky needs in table 2.2. are translated into specific customer profiles. The management wants PsychiatryMatters.MD to be effective and the budget spent well. Therefore the management reckons with four customer groups [8]:
1. Dog customers – These customers are the users that Organon has to lose. Dog customers do not influence sales. An example is a consumer.
2. Satisfied customers – These customers can influence sales indirectly. An example is a consumer under treatment of a psychiatrist and using the CD&T database. Another example is a non-prescribing medical professional, not Organon-minded and/or no special interest in psychiatry.
3. The potential core customers – These customers are part of the target group. An example is a medical professional that does not prescribe Organon products. Organon has permission to contact this customer only by e-mail. Other customer characteristics are: average turnover, habitant of G-10 country (country in top 10 sales), visits PsychiatryMatters.MD at least twice a month, mainly user of drug database. In the future most of these customers will increase sales.
4. The star or core customers – These customers are high prescribers of Organon products. Other customer characteristics are: Organon minded; an ambassador; permission to contact (multi channel); high (potential) turnover; psychiatrist; habitant of G-10 country; visits PsychiatryMatters.MD at least twice a week; loyal and participates in forum and uses his library often. These customers deserve all attention. Most of the time this type of customers make 80% of sales.
Figure 2.3. indicates the marketing funnel. To collect the customers the funnel method is used. The marketing funnel is a continuous flow from the public all the way to Organon’s best customers. Each stage in the funnel is much smaller in terms of numbers of people than the previous stage. The funnel starts to attract internet surfers looking for information on psychiatry. In this first stage the characteristic origination of the target group is very important. This first stage of the research determines this origination. In the first stages, Internet users are not heavy users of PsychiatryMatters.MD. The users are aware of the presence of PsychiatryMatters.MD at the Internet. This stage contains dog customers and the target group. The last stage contains the target group that is loyal users of PsychiatryMatters.MD and other Organon product based sites. With the funnel-methodology, Organon wants to make of all medical professionals within the target group prescribers.
Actual data of the customer behavior and use of PsychiatryMatters.MD is necessary to measure the presence of the target group. A continuous analysis and interpretation of the data will enhance understanding this customer behavior. It is important with this analysis to match the behavior with the business objectives set at the release of the web site. The result of the analysis is intelligence, which enables differentiation of the content and services. A continuous feed back loop must ensure the effectiveness of the funnel method. The next section explains all about the use of marketing intelligence in the organization of Organon.
Figure 2.3. The funnel
FUNNEL Disease sites Brand sites Awareness Diagnosis Choice Compliance Loyalty PsychiatryMatters.MD
2.3. What is marketing intelligence?
The definition of marketing used in this research is as a compilation of activities. Marketing activities are those activities that are based on the exchange of values between Organon and one or more pressure groups in the external surroundings of Organon. These activities are based on information and they must help fulfill objectives set by Organon [10].
With this definition, it is obvious that it is about the relation between organization and surroundings in the marketing decision-making process, in this case especially the relation of Organon with its online prospects. The definition also states that the decision-making process is based on accurate intelligence. In this research the intelligence concerns data of the behavior of the user group of PsychiatryMatters.MD. So largely this data concerns marketing intelligence.
Figure 2.4. indicates the decision making process. This section uses the three-step decision making process [7]. These steps make clear the importance of market intelligence in marketing strategy. Market intelligence is the foundation of every marketing strategy.
In the first step, intelligence, the market is screened for signals that need a response or a decision. These signals are the first symptoms of a problem and the input of a decision making process. In this stage the borders of the problem are set, the description of the problem. In addition, in this first stage the problem owner is identified. The analysis in this thesis largely is a comparable screening of an online target group of PsychiatryMatters.MD.
The design step focuses on the development of possible campaign solutions adapted to the new market situation.
Intelligence
Design
Choice
The last step of the decision making process, chooses a specific e-campaign from the in previous step developed alternatives. This is the optimal alternative.
If decision-making is strongly structured (programmable decision-making) it is possible to leave it to the information system. However, with weakly structured decision-making and this is the case with the marketing at Organon it is not possible. The external influences of the psychiatry market will continuously change and the market with it. Nevertheless, the information systems are very helpful. Information systems have the role of decision-support instead of decision-making. The possible support differs in each step of the decision-making process and because of that the information systems differ also in every step. The decision making process distinguishes three supports functions [7]:
Monitoring the online prospects; in an online market where people subscribe and leave cookies it is possible to trace users accurately.
Analysis of the data extracted during the monitoring process; all the data sources will be mixed, related and mined to create the intelligence on which the Internet marketing management can base their decisions. The Internet marketing management uses the intelligence to adapt the online campaigns. Of course complete the marketing management could benefit of this intelligence.
Use of models, in this research no models are used.
Because this research uses no models, only a comparison with a data oriented Marketing Decision Support System (MDSS) is possible [7]. The usefulness in this comparison is the advantages and disadvantages it makes clear. Figure 2.5. indicates several marketing management support systems. This figure shows that a further development of the marketing system at Organon is possible. In this research the analysis path is comparable to a data-oriented MDSS, so only the first three shall explained extensively the following four not.
6. Model oriented MDSS with causality
4. Model orientated MDSS with definition model 5. Model oriented MDSS without causality 7. Marketing expert system
3. Data orientated MDSS 2. Marketing report system 1. Non-marketing report system
1. Non-marketing report system;
The biggest part of these systems is developed for use by other departments but as by-product they can produce useful marketing information. A nice example is order pick systems, the marketing departments receives a resume of the sales in a period. Disadvantages of these systems are the disability to create specialized reports of products or regions.
2. Marketing-report systems (MRS);
For fulfilling, the requirement of marketing it is necessary to save detailed data during n time periods. Marketing-report systems report separately for consumers. In addition, these systems contain historical data for comparing.
3. Data-oriented Marketing Decision Support Systems;
A major disadvantage of MRS is that overviews are not possible. Data-oriented MDSS are flexible systems that enable easy data configuration. It is possible to create several overviews and configure the customer data in easy interpretable charts. In contradiction to MRS, data-oriented MDSS extract data from a number of sources.
Summarizing the advantages and of course if the data collection is reliable the MDSS delivers the next four important groups of market intelligence, indicated in the table 2.6. The intelligence that is collected during this analysis of customer behavior at PsychiatryMatters.MD is thickly outlined.
External Internal
Status intelligence
This is information about the empirical situation. This information concerns market research.
Information departments, such as: sales info from Marketing, non-payment info from Financials, stocks and returning info from Logistics, etc.
Trend intelligence
Concerning the same subjects as external status info, but now it is compared during time periods.
Concerning the same subjects as internal status info, but now it is compared during time periods.
Comparing intelligence
To label a signal in the market as problem or opportunity, the data has to be compared with market means.
To label a signal in the market as problem or opportunity, the data has to be compared with standards, management objectives, budgets, etc. Forecast
intelligence
Problem recognition gathered from historical data entails in late response. Forecasting could lead to competitive advantages.
An explanation of the four groups of intelligence that this research uses is:
External status intelligence: information about characteristics of the target group will
be defined in the analysis of this research. Also clusters with homogeneous characteristics will be defined.
External trend intelligence: the clusters will be tracked during four successive
months.
External comparing intelligence: for determining characteristics the mean from the
target group will be compared with the means in the whole customer group.
Internal comparing intelligence: the results will be compared with the management
objectives.
2.4. Summarizing the chapter
In this chapter, the theoretical framework for this research is set. This theoretical framework is the foundation of this research and report.
Section 2.1. starts of to define PsychiatryMatters.MD as a commercial virtual community. With this the definition for its most important feature is explained, namely the retention of users. This research uses for the retention of users the term ‘stickiness’. Drivers of stickiness cause a CVC to be sticky for its target group.
Section 2.2. explains the customer group. The target group is all medical professionals with an interest in psychiatry. PsychiatryMatters.MD has to fulfill all the information needs of the target group to be sticky. This section also explains the funnel method to catch the target group for the CVC.
Section 2.3. explains another advantage of a commercial virtual community is described. A CVC makes it possible to extract data about the customers from the Internet. This section defines marketing intelligence and its features.
Chapter 3 - The environment of PsychiatryMatters.MD
This chapter describes the three elements in the environment that influence CVC PsychiatryMatters.MD, as depicted in figure 1.2. of section 1.5. To understand the purpose of this CVC the organization and the objectives behind the portal will be described.
The first section 3.1. describes the first element, namely the organizational structure of Organon. From the companies history a good definition of its core competencies can be made. And logical results of the companies’ core competencies are its main markets.
The second section 3.2. describes the second element, namely the business objectives for PsychiatryMatters.MD. These goals will be compared with the results of this research.
The third section 3.3. describes the third and most important element, namely the customer group.
3.1. Organon NV
This section 3.1. explains the first element in the environment that influences PsychiatryMatters.MD, namely the Organon e-business department. Akzo Nobel’s Pharma Group comprises three healthcare businesses whose activities serve customers across the globe. Figure 3.1. indicates the organizational structure of Akzo Nobel. Organon is an international player in several key areas of human pharmaceuticals, in particular prescription medicines.
Organon Intervet
Intervet is an animal healthcare company, supplying customers with animal vaccines and pharmaceutical products. Diosynth is a producer of complex active biopharmaceutical ingredients for the pharmaceutical sector.
Organon is a Dutch research-based pharmaceutical company and the largest pharmaceutical business unit of Akzo Nobel. The company is active in more than 100 countries with its own affiliates in 55 countries. Major product groups are oral contraceptives, infertility treatments, preparations for menopausal complaints, depression and psychosis, and muscle relaxants. Product groups involved in this research are depression and psychosis. Organon discovers, develops, manufactures and markets products in the following areas:
• Reproductive medicine; • Contraception; • Infertility; • Menopause; • Atherothrombosis; • Immunology; • Rheumatoid arthritis; • Schizophrenia; • Depression.
The last two areas are involved in this research.
A short description of Organon’s activities is research and development of innovative medicines to improve human health. Organon's mission statement is: to be a reliable and respected partner of health care professionals by providing products and services for the world-wide improvement of human health and quality of life.
Worldwide more than 4000 company personnel are involved in marketing within Organon. This huge number reflects the place of information services in the modern pharmaceutical industry and the commitment now needed to present new pharmaceutical products to healthcare professionals. A completely new movement in this marketing of pharmaceutical products is the use of the Internet to enhance contact with the prospects of Organon. This research explores if it is a successful strategy, as it already proved to be for other companies. Figure 3.2. on the next page indicates the organizational structure of Organon. The right branch of the organizational tree indicates the marketing department with e-business. The next section explains more about the e-business department.
3.1.1. E-business department
This research is conducted by order of the e-business department that is part of Organon marketing. Figure 3.3. depicts the structure of Organon e-business. The organizational tree shows clearly the combination of marketing and technical operations. The definition of business in this research is the same as the e-business department at Organon uses, namely:
‘The pervasive use of Internet-enabled technologies to interact, collaborate and transact business with customers, suppliers, partners, employees and shareholders’. Price Coordination Marketing & Operations International Production & Quality Affairs Medical Affairs Research & Development International Pharma Policy eBusiness International Marketing Staff Departments Organon Management
Figure 3.2. Structure Organon
Secretary
eBrand-Manager (Consumer focus) eBrand-Manager (HCP focus)
Medical Advisor eMarketing Manager
Database Marketer Market Research Analyst
Web Engineer
Webmaster
Internet Strategy Manager eIntelligence & Knowledge Management eBusiness Director
The e-business department has two major objectives:
1) The e-business department wants to create via the Internet mutually beneficial relationships with Organon’s clients. Loyal relationships are built up (physicians and consumers to start with) by offering the customers high quality information related to health issues and products. PsychiatryMatters.MD has to fulfill the information needs of the customers. The e-business department develops most of its work in this area of Internet marketing.
2) The e-business department simulates and co-ordinates the use of Internet technologies in all processes cross the company wherever it makes business-sense.
The core activity is customer relation management. The clear objective for CRM is to enable a customer to interact with a company through various means including the Web, PDA, e-mail and receive a consistent level of quality service. The e-business department wants to be an e-knowledge center and fully reap the benefits that Internet offers for Organon in both Consumer (DTC) and Direct-To-Physician (DTP). PsychiatryMatters.MD is a DTP initiative.
3.2. The business objectives
This section 3.2 explains the second element in the environment that influences PsychiatryMatters.MD, namely the business objectives that guide PsychiatryMatters.MD. This research checks if the users and their behavior stroke with the objectives set for PsychiatryMatters.MD. In this section, these business objectives of Organon will be explained. After a start with the strategic objectives in section 3.2.1. the internet marketing objectives will be described section 3.2.2. Finally section 3.2.3. describes the objectives of PsychiatryMatters.MD.
3.2.1. Strategic business objectives
A clear objective for the whole Organon-company is set, namely: ‘increasing sales and market share for its therapeutic products. This is accomplished by marketing
newer, better medications to physicians, which lead to increased prescriptions, thus ultimately increasing drug sales.
Figure 3.4. indicates the integration of an Internet marketing strategy into business and general marketing strategies. The CVC PsychiatryMatters.MD only has added value if the goals of this site stroke with Organon’s corporate image and purposes [12]. The internal influences include corporate objectives and strategy. These in turn will be among the influences on marketing strategy that directly influences the Internet marketing strategy.
Organon uses a clearly defined Internet marketing strategy. Organon wants to create mutually beneficial relationships with customers via the Internet. A clear Internet marketing strategy is necessary. First because the Internet marketing strategy is part of the broader strategic marketing planning process. Second because the Internet is a significant communications channel towards its customers. A clear Internet strategy supports the main thrusts of the marketing and business strategy.
Organon’s corporate objectives and strategy Organon’s marketing strategy Market structure and demand Competitor Strategy Emerging opportunities and threats Internal influences External influences Organon’s Internet marketing strategy
3.2.2. Organon’s Internet marketing objectives
Product launches have become more prevalent in recent years, as Organon, wary of competing drugs, seeks market penetration as fast as possible. The Internet is a new channel to contact physicians. In this channel a large number of physicians will emerge that receive drug information via the Internet. Internet marketing has the potential to enhance sales for Organon. The e-business department wants to create via the Internet mutually beneficial relationships with Organon’s clients. Therefore web based customer analysis and profiling is applied to define customer needs and drivers. The incessant goal for the e-business management is to use the Internet for educating physicians on newer, better medications, which ultimately leads to increased drug sales.
3.2.3. Objectives PsychiatryMatters.MD
The corporate image is used as the framework to position PsychiatryMatters.MD and its services. Organon uses a brand and a proposition that is well established in the offline world and they can build on the brand by duplicating it online. The only risk of migrating existing brands online is that the brand equity may be reduced if the site is of poor quality. PsychiatryMatters.MD carries the brand of Organon on every site. Important is that PsychiatryMatters.MD holds on to the proposition Organon uses. The proposition of Organon is “Health Matters” and the created portal is called PsychiatryMatters.MD. A clear Internet value proposition helps distinguish the site from its competitors. It also helps provide a focus to marketing efforts. The proposition links the CVC to normal product propositions of a company or its product. Organon wants PsychiatryMatters.MD to be the “essential psychiatry resource”. PsychiatryMatters.MD is positioned as the most unbiased among professionals, anytime anywhere and a one-step-shop. In the positioning of PsychiatryMatters.MD, Organon e-business uses the core values: Practical, Independent, Professional and Up-to-date.
Now the company is able to differentiate the CVC from similar competing services. Its distinguishing features in online and offline promotion differentiate the CVC from those of its rivals. The keyword while creating and upgrading PsychiatryMatters.MD is customer orientation, providing content and services on a web site consistent with
the different characteristics of the audience of the site. Important for PsychiatryMatters.MD success is the size of Organon’s online target group. It must target customers who can have the biggest impact on the company’s profitability. The target group of PsychiatryMatters.MD is all medical professionals with interests in psychiatry. The next section defines the online customer group and target group.
3.3. Online customer group
Psychiatry is one of the fastest growing areas for Organon. Regarding PsychiatryMatters.MD Organon has explicitly stated to focus strongly on psychiatry. The largest lever for raising product sales is to improve the influence on medical professionals with an interest in psychiatry. With PsychiatryMatters.MD now Organon is using the Internet channel to contact the customer group. This section explains this third element on the environment, namely the online customer group. Not all customers of Organon are users of PsychiatryMatters.MD. In addition, not all the users of PsychiatryMatters.MD are customers of Organon. It is important to know what the reach is of the analyses in the research. In statistic terms, this section defines the online population of the analyses. Important in the definition of the population are the prospects and the suspects. Prospects are considered registered users of PsychiatryMatters.MD, but with whom or which no sales transaction is pulled of. Suspects are considered individuals or companies in the target group, but with whom or which no relationship exists. Figure 3.5. defines the population:
A. The ‘offline’ existing relationships of Organon;
B. Registered users of PsychiatryMatters.MD. The registered users could be considered as prospects for the ‘offline’ Organon organization;
C. Core customers; Registered users of PsychiatryMatters.MD and relationships of Organon (A+B); D. Suspects. Core C Prospects Organon B A Suspects D
Existing relationships of Organon
Registered users of PsychiatryMatters. MD
Within the online customer group (B+C), web based customer behavior analysis is possible. If a differentiation of strategy is successful, more prospects have to shift slowly from the b-box into the c-box, the core customers. Before customers turn into core customers, they should become profound users of PsychiatryMatters.MD this is accomplished by differentiating and personalizing the web content and services. The online target group is only a part of the total population, especially in countries with low Internet penetration. Table 3.6. shows the quantification of the online target group.
Online market size (in 1000)
G10 area Online psychiatrists Online total target group
US 37 537 Europe 20 442 Rest G10* 10 130 Total G10 67 1109 Total SI** 6 433 Total G10 + SI 72 1542
*G10 are the top 10 sales countries for Organon ** SI are other countries than G10
Source: Marketing plan PsychiatryMatters.MD by VODW Marketing
3.4. Summarizing the chapter
In this chapter, the environment of and the influences on PsychiatryMatters.MD are described as indicated in figure 1.2. of section 1.5. This research draft distinguishes three elements that directly influence PsychiatryMatters.MD. These elements are the e-business department at Organon, the business objectives and the customer group. The first element of influence is the e-business department producer behind PsychiatryMatters.MD described. .
The second element of influence are the objectives of PsychiatryMatters.MD that are explained in the second section. Very important are the strategic objectives used as a framework for PsychiatryMatters.MD.
The third element is the online market that is shortly described in the third section. Four groups are distinguished in the target group: the core-customers, the prospects, the suspects and the offline relationships of Organon. This online target group is quantified for the top-10 sales countries of Organon, namely more than 1.5 million medical professionals are online.
Chapter 4 - Design of the research
This chapter describes the design of the online analysis. This chapter uses the same the two-stage partition as mentioned before in chapter 1. Figure 4.1 indicates this two-stage partition. The first stage, 'origination' depicts the analysis of the use of engines and search strings by the target group. This stage regards analysis of characteristics before the user logs on to PsychiatryMatters.MD. The second half, 'stickiness and clustering' depicts the analysis of the stickiness and characteristic preference for services by the target group. This stage involves analysis of characteristics after the user logs on to PsychiatryMatters.MD. In the second stage, clustering is the method of grouping the users with the same preference. The sections regarding the two stages explain the need for data, the collection of data and processing of the data.
Section 4.1. describes the data sources and the data mining process. Section 4.2. describes the first stage that will define the characteristic origination of the users. Section 4.3. describes the second stage that defines several characteristic segments.
4.1. In coming data sources
This first section describes the data sources. Regarding the data, this research is dependent of two main factors. First, it is all about the kind of information we use. The data we use in this analysis concerns personal profiles and personal behavior of consumers and medical professionals. By law, these personal profiles are limited in use for corporate intentions. Still now the legislation about this topic is developing. The technology is developing very fast, so this legislation has to be adapted all the time. [15, Law on protection of personal data files (WBP)]. This law does not allow
Log on
1. Origination 2. Stickiness and clustering
N C P N C P
Figure 4.1. Research process
N = Need C = Collection P = Processing
using personal profiles without permission of the owner. The user of PsychiatryMatters.MD permits Organon to use the e-mail address for some specific information regarding psychiatry.
Second, a part of the data this research uses is extracted from the web log files. Therefore, this research is dependent on the state of technology. If the server of PsychiatryMatters.MD is down, the server produces no web log files and no user data during this period is saved.
The log files register the web behavior of the customers. For everyday, the server saves a log file of PsychiatryMatters.MD. These web log files are created with cookies.
Cookies are small computer files that store information about computer users (with protection for confidential or sensitive information) to speed up and otherwise improve the user’s on-line experience. The purpose of cookies is to identify and track users in order to prepare customized WebPages such as on-line shopping carts. These cookies allow users to register only once instead of each time they access a site. Cookies may track information such as pages visited, operating system, browser, and information enter a form. [24]
For taking random samples a tight population framework is needed [1]. This framework must equal the target population and customer group in this research to be representative for the whole target group of ‘PsychiatryMatters.MD’. Cover errors are not allowed. The registration user table, hereafter named as user table, is suited as population framework. A user that is subscribed at PsychiatryMatters.MD has a user identification number. This number labels the number every time it logs on to PsychiatryMatters.MD and starts using its services. Through this method, all the actions in the user's behavior can be linked to a profile that belongs to this user. The user makes this customer profile him or herself when subscribing at PsychiatryMatters.MD. The user table saves the customer profiles. A customer profile contains six variables. Window 11.1. in the site map shows the customer profile made by a user. Figure 4.2. on the next page depicts a piece of the user table with six user profiles. In the user table, the target group is easily identified. The six variables are:
• First Name; • Profession;
• Last Name; • Country;
User ID User name First Name Last Name Profession code Email address Country code
29272 abruceclark bruce clark -99 bclark@………… 163
29271 kwquirky kenneth wilkerson 11 kwquirky@………. 163
29270 joango Joan Goldman -99 sgold9467@…………. 163
29269 Rachelle Rachelle Mccoy 10 mccoy@………….. 163
29268 PSYCHOLOGIST4897 Rizwan Tareef -99 RIZWANTARIF@…… 119
Log files save the actions in the user's behavior at the CVC. The server of PsychiatryMatters.MD saves a log file every day. A log file saves every click at a site made by the user.
A data mining process connects the user table and the log files through the user identification number.
4.1.1. Data mining
Figure 4.3. indicates schematically the data mining process. Three stages achieve data quality, namely [21]:
1. Cleansing; 2. Matching; 3. Consolidation.
The data in the user table is al ready in a convenient order. In contrary, the log files are not yet conveniently ordered and not yet in easily usable tables. The data cleansing stage parses, corrects, standardizes and enhances the data in the log files for accurate matching. Parsing is the first critical component in data cleansing. This Figure 4.2. The user table
process locates, identifies and isolates the individual data elements in the log files. For example, components include such data as the user identification number, user's profession, search strings and search engines.
Correction is the second critical component in data cleansing. If the data comes from a variety of sources, problems can appear:
• Variations in abbreviations, formats, etc., because of individual preferences of the person entering the information;
• Outdated information due to name and address changes;
• Transpositions resulting from keying errors;
• Misspellings.
The next stage is standardization, arranging customer information into preferred and consistent tables. Important in this stage is the removal of all Organon employees for an objective result.
Enhancement, the final step in the data cleansing stage, appends new data and completes missing information.
The matching stage identifies similar data within and across data sources. In this stage, comparisons are made within and across the user table and the log files to locate similar information. Using the cleansed information and match standards duplicates can be eliminated. Finally, the matching data elements are consolidated and placed into a database or a data warehouse. In this data warehouse, the analysis of the data is possible and creating knowledge is the next step.
4.2. The user table as population framework
Section 4.1. introduces the user table as the population framework for this research. The population framework must be reliable for this research to representative. This section explains the method that is used to check the user profiles. The user table contains all the user profiles of the users of PsychiatryMatters.MD. The users when subscribing at PsychiatryMatters.MD create the customer profiles. Therefore the user profiles in the user table must be checked on their existence for the user table to be a reliable population framework.
A quick and easy method to check the user profiles would be to use email addresses. However, as stated in section 4.1. the personal profiles are limited in use by Organon. Unfortunately Organon is not allowed to use the email addresses from the customer profiles for a direct mail interview with the psychiatrists.
The method to do so is not complicated. A sample consisting only of Dutch psychiatrists will be looked for in the Dutch phonebook (www.telefoongids.nl) to check their existence. This method gives an indication on the reliability of the user table as the population framework in this research.
4.3. Origination of the user
This section explains the determination of the origination of the users. The first stage determines the characteristic origination of the target group compared to the whole customer group. The result of this analysis is the characteristic origination of the target group for the portal ‘PsychiatryMatters.MD’. Due to this analysis, a better positioning of the portal at the World Wide Web is possible.
It is possible to measure two distinct characteristics in the origination, namely the search strings used and the search engines used. Examples of search engines are 'www.google.com' or 'www.search.com'. Internet surfers use specific expressions, the search strings to collect more information at the World Wide Web about a specific topic The first stage uses the profile analysis to determine characteristic search engines and -strings. Figure 4.3. indicates the formula of the profile analysis. If the index calculated for a search string or -engine is higher than one hundred, this search string or -engine is considered characteristic [7].
P - value of target group - this value is the proportion of a certain characteristic within
the target group. This analysis determines the characteristic regarding the use of a search engine or a search string.
P value of all users - this value is the proportion of a certain characteristic within the
customer group. This analysis determines the characteristic regarding the use of a search engine or a search string.
Index – this index reflects the relation of P value of target group to the P value of all
users. An index bigger than one hundred suggests a differentiating characteristic. The P-values for the whole population are estimated in a random sample survey [1]. Analysis of all search actions by the target group through the existence of PsychiatryMatters.MD is too time consuming. In addition, the calculation capacity of the computer would be too small.
Figure 4.5. Formula for index in profiling analysis
4.3.1. Method of sample taking
In this case, the survey is a binomial distribution [1]. This means each trial has two possible outcomes, identified as certain outcome happening or certain outcome not happening. In this case more specific it concerns the use of a search string or the use of search engine. Thus visitors can use a specific string and engine or they do not use it.
This research uses the log-files of November. In time of conducting this analysis, this month was the most recent data available. For this analysis an accuracy of 5% and a reliability of 99,7% is demanded. Important for meeting these demands is the sample size. Because the population standard deviation is not known in front, these demands can only be determined afterwards. The formula in figure 4.4. is used for checking if the samples are representative.
4.4. Clustering the users
The goal is to determine if the target group has a characteristic preference for certain services at the CVC. Probably a psychiatrist will use other services at the CVC than a consumer. The objective of the cluster analysis is to divide the customer group into several groups within each group or segment homogeneous characteristics. Each user must have more similarities with its family users than with users out other segments. Cluster analysis is often used for market segmentation, like this case. Freely translated the users with same characteristic service preference will be
With n is sample size, z is z-value belonging to reliability, e = accuracy, p is likelihood of a certain outcome happening and q = 1-p is likelihood of a certain outcome not happening
clustered in one group. The characteristics on which this clustering is based are the active variables.
Active cluster variables i.e. service preference
1. About this site 22. Links: societies 2. Add notes 23. Links: support groups 3. Add to favourites 24. Links: web alerts 4. Add to library 25. Literature: alerts 5. APA 26. Literature: alerts archive 6. CD&T: search 27. Literature: case studies 7. D&T: specific item 28. Literature: review 8. Conference: calendar 29. Literature: specific item 9. Conference: reports 30. Literature: trial reports 10. Conferences: specific item 31. Magazine: archive 11. Disclaimer 32. Magazine: latest issue 12. DrugDB: search 33. Mail to a colleague 13. DrugDB: specific drug 34. My profile: edit 14. Feedback 35. News: specific item 15. File demo: full demo 36. Print
16. File demo: specific 37. Privacy policy 17. Forgotten password 38. Resources: CD&T 18. Help 39. Resources: MyLibrary 19. Links: academic groups 40. Resources: News 20. Links: journals 41. Resources: PubMed 21. Links: others 42. Search: advanced
The active variables are the services the users choose at PsychiatryMatters.MD. [22] For determining the clusters in this case one active variable or characteristic is used, thus it is a monotheistic measurement. For example, we could classify people solely based on their gender (a monotheistic classification), alternatively we could use gender and blood group (a polytheistic classification). Table 4.7. indicates the services possible to use at PsychiatryMatters.MD. In the final segments, users are clustered with a familiar preference for services. The clustering is based on the service preference. After the clustering process, characteristics other than the active variable name the clusters. These characteristics are the passive (background) variables. The passive variables are background variables that do not influence the cluster process. For the inactive variables, the frequencies are calculated.
The background variables are elements from the subscriber forms. Unfortunately, these forms do not contain unlimited data. Usually these description forms do scare users away for subscribing at a portal. In total 6 different characteristics variables will describe or name the clusters. Namely one active variable:
Number of resources – the number of services at PsychiatryMatters.MD that are
used by a cluster.
The five descriptive variables are:
Average Use intensity (AUI) – This is the average number of actions in one session
of one user. Thus, sum of the sessions divided by number of actions in log files. One session is as one user log on at PsychiatryMatters.MD. An action is as one user in a session chooses for a certain service at PsychiatryMatters.MD. This variable is the equivalent for the stickiness. Stickiness is defined as in increased frequency of user’s revisits. The duration is translated into number of actions in one session. The revisits are counted as new sessions.
Proportion of prescribers – the share of prescribers in the total number of users in
this cluster at PsychiatryMatters.MD.
Proportion of medical professionals - the share of medical professionals in the total
number of users in this cluster at PsychiatryMatters.MD.
Countries G10 – the top 5 countries of origination from the G10
Service preference – top 4 of service preference in this cluster at
PsychiatryMatters.MD
In this cluster analysis, an agglomerate method is applied. A divisive method begins with all users in one cluster. This cluster is gradually broken down into smaller and smaller clusters. Agglomerate techniques start with (usually) single member clusters. These are gradually fused until one large cluster is formed.
4.4.1. Cluster algorithm
This analysis uses a hierarchical cluster process to create an initial cluster solution. Hierarchical means that the resultant classification has an increasing number of nested classes. For this cluster process the algorithm of Ward is used because it is proven to return the most reliable results. Ward cluster algorithms can only be used with Euclidean metric distance definition [19]. Figure 4.8. on the next page explains the Euclidean metrics.
The Ward algorithm calculates the means of all variables in each cluster. The next step is to calculate for each case its variables the squared Euclidean distance with the cluster mean. Then the distances of al cases are summed. By each fusion step the clusters with the smallest difference between the total sum of the squared Euclidean distances are combined. In this process the cluster tree or dendrogram is created. The dendrogram shows all the clusters containing the elements. The dendrogram is used to choose the initial solution for the number of clusters [12, 15]. After the clusters are determined the clusters will be tracked for the four successive months, August 2001, September 2001, October 2001 and November 2001. For every month, the values for the six variables are calculated. In this way, developments in the characteristics of the clusters can be monitored. The stickiness is determined during these four months.
4.5. Goals that will be measured
This section translates the objectives described in chapter 3 into goals that will be compared with the results of the online analysis. This comparison will show a nice overview of the accomplishments so far.
1. Create a CVC for online services and for offline use (database; CRM).
PsychiatryMatters.MD must be a CVC that offers services that attract users. The services at PsychiatryMatters.MD must be suited for online and offline use.
Euclidean Metrics
These measure true straight-line distances in Euclidean space. Note that the distance from Manchester to Chicago would not be Euclidean unless you bored a tunnel though the earth. The flight path of plane would follow the curvature and hence would not be straight-line distance.
In an univariate example, the Euclidean distance between two values is the arithmetic difference, i.e. value 1 – value 2. In the bivariate case, the minimum distance is the hypotenuse of a triangle formed from the points. For three variables, the hypotenuse will be extended through 3-variables space. Although difficult to visualize, an extension of the Pythagoras theorem will give the distance between two points in n-dimensional space.
